drone-audio-detection-05-17-trial-2

This model is a fine-tuned version of MIT/ast-finetuned-audioset-10-10-0.4593 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0593
• Accuracy: 0.986
• Precision: 0.9917
• Recall: 0.9904
• F1: 0.9910

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0004904803475016782
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Use OptimizerNames.ADAFACTOR and the args are: No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.01632118285229547
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
0.2162	1.0	250	0.1438	0.948	0.9980	0.9354	0.9657
0.0729	2.0	500	0.0963	0.974	0.9967	0.9699	0.9831
0.1047	3.0	750	0.0790	0.9755	0.9974	0.9712	0.9841
0.0884	4.0	1000	0.0551	0.984	0.9929	0.9866	0.9897
0.0459	5.0	1250	0.0593	0.986	0.9917	0.9904	0.9910

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.1